Awesome, will take a look!

I'm fixing the lint etc.

LMK what you think about the approach to conversion; it's kind of fiddly.

@U45T93RA6 may be interested in that

ah, a possibly-controversial point I just noticed. :style/indent is nominally only supported for macros, and the PR respects that. but cljfmt doesn't care and there are lots of useful :style/indents in the wild on defns too. one example is Toucan, to get nice

(db/update! SomeModel id
  :field 7
  :other_field "bar")

I'm ok supporting it for functions, I can't see how one would add a :style/indent accidentally if not for that purpose

agree. my only concern is that they're not namespaced 😞 cljfmt is too naive to understand an indent spec for 'toucan.db/insert! should be used when (:require [toucan.db :as db]) (db/insert! ...)

yeah, good point

if the classpath has 2 functions of different ns with same name and both with style/indent, one of them would use the wrong style/indent right

yeah, that's correct.

the Right Answer is probably to make cljfmt smart enough to understand ns aliases, then it would work unless you :refer or :use.

yes, so maybe we should start with only macros support for now?

but would be nice to address a issue on cljfmt about that

SGTM. I'll look into that as a PR today too; hopefully it's not too hard.

how about I future-proof the PR by making it use macros unqualified, and functions qualified?

hmm, since cljfmt doesn't support that and we don't even know if is easy or even possible I wouldn't suggest do this premature optimization, we could easily change that later after cljfmt changes

fair.

https://github.com/weavejester/cljfmt/pull/291 fixes the aliases on the cljfmt side

Thanks for the ping! Awesome stuff. Stupid q, is a dynamic require like a top-level (require '[foo.bar :as f]) accounted for?

AFAICS it should work for that as well

@UCY0GT0QM I'm about to release a new clojure-lsp version tomorrow, but I'd like to leave this PR for the next one if you don't mind since I'd like to test it in the master a little bit more regarding perf/any issues we may face

sounds good. I can use my local build in the meantime anyway.

Or use a nightly one from #C032YH7P0R2 (compiled with graal) Sounds good, LMK if you will address vemv suggestions or if it's read to one last check

@UKFSJSM38 thoughts on just including the non-macros, with namespacing? it's a tiny change to LSP. we can avoid the need for the cljfmt fix in most cases with (update cljfmt-config :alias-map not-empty), since it's nearly always {}.

Yeah, sounds like it's worth, go ahead

new version up